00:00:33 AI的“一根筋”难题：如何跳出思维的隧道？

00:05:08 AI的“高情商”陷阱：当它说得越流利，你可能越要小心

00:08:53 如何让AI学会“开窍”？

00:13:49 给你的AI，装上一个“立体思维”引擎

00:18:41 AI偷懒的智慧：如何用更少的力气，办更聪明的事？

本期介绍的几篇论文：

[LG] ParaThinker: Native Parallel Thinking as a New Paradigm to Scale LLM Test-time Compute  

[Tsinghua University]  

https://arxiv.org/abs/2509.04475 

---

[CL] Knowledge Collapse in LLMs: When Fluency Survives but Facts Fail under Recursive Synthetic Training  

[Holistic AI & University College London]  

https://arxiv.org/abs/2509.04796  

---

[LG] Bootstrapping Task Spaces for Self-Improvement  

[Meta Superintelligence Labs]  

https://arxiv.org/abs/2509.04575  

---

[CL] Enhancing Diversity in Large Language Models via Determinantal Point Processes  

[Boston University & University of Maryland & University College London]  

https://arxiv.org/abs/2509.04784  

---

[LG] Recurrent State Encoders for Efficient Neural Combinatorial Optimization  

[University of Hildesheim]  

https://arxiv.org/abs/2509.05084